Show the code
#|
if (get_count) {
oa_count$oa <- openalexR::oa_fetch(
entity = "works",
search = "",
output = "list",
count_only = TRUE,
verbose = TRUE
)$count
}Report Assessment Ch2 Technology Visions
Data Management Report
Abstract
A short description what this is about. This is not a tracditional abstract, but rather something else …
Working Title
IPBES_TCA_Ch2_technology
Code repo
Build No: 152
%The BuidNo is automatically increased by one each time the report is rendered. It is used to indicate different renderings when the version stays the same%.
Introduction
All searches are done on all works in OpenAlex. The search in the TCA Corpus is not possibly at the moment, but we are working on it.
The following steps will be done in documented in this report:
Methods
From OpenAlex directly
The search terms are based on the shared google doc. They are cleaned up for the usage in OpenAlex.
OpenAlex Complete
Show the code
#|
if (get_count) {
oa_count$oa_years <- openalexR::oa_fetch(
group_by = "publication_year",
output = "dataframe",
verbose = TRUE
) |>
dplyr::mutate(
publication_year = as.integer(as.character(key_display_name)),
key = NULL,
key_display_name = NULL,
p = count / sum(count)
) |>
dplyr::arrange(publication_year) |>
dplyr::mutate(
p_cum = cumsum(p)
) |>
dplyr::select(
publication_year,
everything()
)
}Vision
The search terms is vision Open Alex search.
Show the code
#|
if (get_count) {
oa_count$vision <- openalexR::oa_fetch(
title_and_abstract.search = params$st_vision,
count_only = TRUE,
output = "list",
verbose = TRUE
)$count
}Show the code
#|
if (get_count) {
oa_count$vision_years <- openalexR::oa_fetch(
title_and_abstract.search = params$st_vision,
group_by = "publication_year",
output = "dataframe",
verbose = TRUE
) |>
dplyr::mutate(
publication_year = as.integer(as.character(key_display_name)),
key = NULL,
key_display_name = NULL,
p = count / sum(count)
) |>
dplyr::arrange(publication_year) |>
dplyr::mutate(
p_cum = cumsum(p)
) |>
dplyr::select(
publication_year,
everything()
)
}Technology
The search terms is technology Open Alex search.
Show the code
#|
if (get_count) {
oa_count$technology <- openalexR::oa_fetch(
title_and_abstract.search = compact(params$st_technology),
count_only = TRUE,
output = "list",
verbose = TRUE
)$count
}Show the code
#|
if (get_count) {
oa_count$technology_years <- openalexR::oa_fetch(
title_and_abstract.search = compact(params$st_technology),
group_by = "publication_year",
output = "dataframe",
verbose = TRUE
) |>
dplyr::mutate(
publication_year = as.integer(as.character(key_display_name)),
key = NULL,
key_display_name = NULL,
p = count / sum(count)
) |>
dplyr::arrange(publication_year) |>
dplyr::mutate(
p_cum = cumsum(p)
) |>
dplyr::select(
publication_year,
everything()
)
}Vision AND technology
Open Alex search.
The search term is vision AND technology
Show the code
#|
if (get_count) {
oa_count$vision_technology <- openalexR::oa_fetch(
title_and_abstract.search = compact(paste0("(", params$st_vision, ") AND (", params$st_technology, ")")),
output = "list",
count_only = TRUE,
verbose = TRUE
)$count
}Show the code
#|
if (get_count) {
oa_count$vision_technology_years <- openalexR::oa_fetch(
title_and_abstract.search = compact(paste0("(", params$st_vision, ") AND (", params$st_technology, ")")),
group_by = "publication_year",
output = "dataframe",
verbose = TRUE
) |>
dplyr::mutate(
publication_year = as.integer(as.character(key_display_name)),
key = NULL,
key_display_name = NULL,
p = count / sum(count)
) |>
dplyr::arrange(publication_year) |>
dplyr::mutate(
p_cum = cumsum(p)
) |>
dplyr::select(
publication_year,
everything()
)
}Show the code
#|
if (get_count) {
oa_count$vision_technology_subfields <- openalexR::oa_query(
title_and_abstract.search = compact(paste0("(", params$st_vision, ") AND (", params$st_technology, ")")),
group_by = "primary_topic.subfield.id",
verbose = TRUE
) |>
openalexR::oa_request() |>
dplyr::bind_rows() |>
dplyr::arrange(key)
## clean up missing or wrong vision_technology_subfields$key_display_name
need_cleaning <- is.na(oa_count$vision_technology_subfields$key_display_name) |
!is.na(as.numeric(oa_count$vision_technology_subfields$key_display_name))
fine <- !need_cleaning
oa_count$vision_technology_subfields <- oa_count$vision_technology_subfields |>
dplyr::filter(fine) |>
dplyr::select(key, key_display_name) |>
dplyr::distinct() |>
merge(y = oa_count$vision_technology_subfields[need_cleaning, -2], by = "key") |>
dplyr::bind_rows(oa_count$vision_technology_subfields[fine, ]) |>
dplyr::group_by(key, key_display_name) |>
dplyr::summarize(count = sum(count))
}Vision AND technology AND marine
Open Alex search.
The search term is vision AND technology
Show the code
#|
if (get_count) {
oa_count$vision_technology_marine <- openalexR::oa_fetch(
title_and_abstract.search = compact(paste0("(", params$st_vision, ") AND (", params$st_technology, ") AND (", params$st_marine, ")")),
output = "list",
count_only = TRUE,
verbose = TRUE
)$count
}Show the code
#|
if (get_count) {
oa_count$vision_technology_marine_years <- openalexR::oa_fetch(
title_and_abstract.search = compact(paste0("(", params$st_vision, ") AND (", params$st_technology, ") AND (", params$st_marine, ")")),
group_by = "publication_year",
output = "dataframe",
verbose = TRUE
) |>
dplyr::mutate(
publication_year = as.integer(as.character(key_display_name)),
key = NULL,
key_display_name = NULL,
p = count / sum(count)
) |>
dplyr::arrange(publication_year) |>
dplyr::mutate(
p_cum = cumsum(p)
) |>
dplyr::select(
publication_year,
everything()
)
}Show the code
#|
if (get_count) {
oa_count$vision_technology_marine_ids <- openalexR::oa_fetch(
title_and_abstract.search = compact(paste0("(", params$st_vision, ") AND (", params$st_technology, ") AND (", params$st_marine, ")")),
verbose = TRUE,
output = "list",
) |>
sapply(
FUN = function(x) {
x$id
}
) |>
unique()
}Open Alex search.
The search term is vision AND technology
Count of Publications
Show the code
# To long search string
if (get_count) {
oa_count$vision_technology_case <- openalexR::oa_fetch(
title_and_abstract.search = compact(paste0("(", params$st_vision, ") AND (", params$st_technology, ") AND (", params$st_case, ")")),
output = "list",
count_only = TRUE,
verbose = TRUE
)$count
}Show the code
#|
if (get_count) {
oa_count$vision_technology_case_years <- openalexR::oa_fetch(
title_and_abstract.search = compact(paste0("(", params$st_vision, ") AND (", params$st_technology, ") AND (", params$st_case, ")")),
group_by = "publication_year",
output = "dataframe",
verbose = TRUE
) |>
dplyr::mutate(
publication_year = as.integer(as.character(key_display_name)),
key = NULL,
key_display_name = NULL,
p = count / sum(count)
) |>
dplyr::arrange(publication_year) |>
dplyr::mutate(
p_cum = cumsum(p)
) |>
dplyr::select(
publication_year,
everything()
)
}Save the counts as data/ch2_technology/oa_count.rds
Show the code
if (get_count) {
saveRDS(oa_count, params$fn_count)
}Download technology AND vision Corpus from OpenAlex
The corpus download will be stored in ch2_technology/pages and the arrow database in data/ch2_technology/corpus_complete. This one will be filtered with the TCA / G;obal Corpus and get the final name data/ch2_technology/corpus.
This is not on github!
The corpus can be read by running corpus_read("data/ch2_technology/corpus") which opens the database so that then it can be fed into a dplyr pipeline. After most dplyr functions, the actual data needs to be collected via collect().
Only then is the actual data read!
Needs to be enabled by setting eval: true in the code block below.
Download Complete Corpus
Show the code
#|
tic()
IPBES.R::corpus_download(
pages_dir = file.path(".", "data", "ch2_technology", "pages"),
title_and_abstract_search = compact(paste0("(", params$st_vision, ") AND (", params$st_technology, ")")),
continue = TRUE,
delete_pages_dir = FALSE,
set_size = 2000,
dry_run = TRUE,
verbose = TRUE,
mc_cores = 6
)
toc()Convert Technology Corpus to Parquet datase
The fields author and topics are serialized in the arrow database and need to be unserialized by using unserialize_arrow() on a dataset containing the two columns.
Show the code
tic()
IPBES.R::corpus_pages_to_arrow(
pages_dir = params$pages_dir,
arrow_dir = params$corpus_complete_dir,
continue = TRUE,
delete_arrow_dir = FALSE,
dry_run = FALSE,
verbose = TRUE,
mc_cores = 3
)
toc()Filter Corpus with TCA Corpus
Show the code
#|
tic()
if (!file.exists(params$fn_ids_tech_in_tca)) {
ids_technology <- IPBES.R::corpus_read(params$corpus_complete) |>
dplyr::select(id) |>
collect() |>
unlist()
ids_tca <- read_corpus(file.path("..", "IPBES_TCA_Corpus", "data", "tca_corpus", "corpus")) |>
dplyr::select(id) |>
collect() |>
unlist()
fn_ids_tech_in_tca <- ids_technology[ids_technology %in% ids_tca]
rm(ids_technology, ids_tca)
saveRDS(ids_tech_in_tca, params$fn_ids_tech_in_tca)
IPBES.R::corpus_filter_ids(
arrow_dir = params$corpus_complete_dir,
arrow_filter_dir = params$corpus_dir,
filter_ids = ids_tech_in_tca
)
rm(ids_tech_in_tca)
}
toc()Check the number of dulicates before running this next block, and then verify the new corpus afterwards. RUN ONLY MANUALY!
Show the code
#|
ONLY RUN MANUALLY!!!!!!!!!!!!!!!!!!!!!!!
(read_corpus(params$corpus_dir) |> group_by(id) |> summarize(n = n()) |> filter(n > 1) |> collect() |> nrow()) / (corpus_read(params$corpus_dir) |> nrow())
years <- IPBES.R::corpus_read(params$corpus_dir) |>
distinct(publication_year) |>
collect() |>
unlist() |>
as.vector() |>
sort()
lapply(
years,
function(y) {
message("\nProcessing year: ", y)
tic()
dataset <- IPBES.R::corpus_read(params$corpus_dir) |>
dplyr::filter(publication_year == y) |>
dplyr::collect() |>
group_by(id) |>
slice_max(
publication_year,
n = 1,
with_ties = FALSE,
na_rm = TRUE
)
# unlink(
# file.path(params$corpus_dir, paste0("publication_year=", y)),
# recursive = TRUE,
# force = TRUE
# )
arrow::write_dataset(
dataset = dataset,
path = paste0(params$corpus_dir, "_deduplicated"),
partitioning = c("publication_year", "set"),
format = "parquet",
existing_data_behavior = "overwrite"
)
toc()
}
)
(read_corpus("./data/ch2_technology/corpus_deduplicated") |> group_by(id) |> summarize(n = n()) |> filter(n > 1) |> collect() |> nrow()) / (corpus_read("./data/ch2_technology/corpus_deduplicated") |> nrow())
rename corpora now
NOW IF EVERYTHING IS OK, DELETE THE OLD CORPUS AND RENAME THE NEW ONE## Export data for sentiment analysis
Show the code
#|
if (!file.exists(params$fn_sent_analysis_parquet)) {
corpus_read(params$corpus_dir) |>
dplyr::select(id, publication_year, ab) |>
arrow::write_parquet(params$fn_sent_analysis_parquet)
}Extract 250 random papers from technology AND vision in TCA Corpus
Show the code
#|
if (!file.exists(params$fn_random_sample_250)) {
set.seed(14)
read_corpus(params$corpus_dir) |>
dplyr::select(id, doi, author_abbr, display_name, ab) |>
dplyr::rename(abstract = ab, title = display_name) |>
dplyr::collect() |>
dplyr::slice_sample(n = 250) |>
dplyr::mutate(
abstract = substr(abstract, start = 1, stop = 5000)
) |>
writexl::write_xlsx(path = params$fn_random_sample_250)
}Publications over time
Show the code
## | |
if (!file.exists(params$fn_publications_over_time)) {
read_corpus(params$corpus_tca_dir) |>
dplyr::select(publication_year) |>
dplyr::arrange(publication_year) |>
dplyr::collect() |>
table() |>
as.data.frame() |>
dplyr::mutate(
publication_year = as.integer(as.character(publication_year)),
p = Freq / sum(Freq),
p_cum = cumsum(p)
) |>
dplyr::rename(
count = Freq
) |>
# oa complete
dplyr::left_join(
x = oa_count$oa_years,
by = "publication_year",
suffix = c("", "_tca")
) |>
# oa vision
dplyr::left_join(
y = oa_count$vision_years,
by = "publication_year",
suffix = c("", "_vision")
) |>
# oa technology
dplyr::left_join(
y = oa_count$technology_years,
by = "publication_year",
suffix = c("", "_technology")
) |>
# oa vision technology
dplyr::left_join(
y = oa_count$vision_technology_years,
by = "publication_year",
suffix = c("", "_technology_vision")
) |>
dplyr::rename(
count_oa = count,
p_oa = p,
p_cum_oa = p_cum
) |>
saveRDS(file = params$fn_publications_over_time)
}Show the code
if (length(list.files(path = dirname(params$fig_publications_over_time), pattern = basename(params$fig_publications_over_time))) < 2) {
sec_axi_fact <- 0.5e-5
figure <- readRDS(params$fn_publications_over_time) |>
dplyr::filter(publication_year >= params$temporal_from) |>
ggplot2::ggplot() +
#
ggplot2::geom_bar(ggplot2::aes(x = publication_year, y = count_tca, fill = "Nmber of publications per year in TCA Corpus"), stat = "identity") +
ggplot2::geom_line(ggplot2::aes(x = publication_year, y = p_cum_oa / sec_axi_fact, color = "Cumulative proportion OA Corpus"), size = 1.5) +
ggplot2::geom_line(ggplot2::aes(x = publication_year, y = p_cum_tca / sec_axi_fact, color = "Cumulative proportion TCA Corpus"), size = 1.5) +
ggplot2::geom_line(ggplot2::aes(x = publication_year, y = p_cum_vision / sec_axi_fact, color = "Cumulative proportion vision only corpus"), size = 1.5) +
ggplot2::geom_line(ggplot2::aes(x = publication_year, y = p_cum_technology / sec_axi_fact, color = "Cumulative proportion technology only corpus"), size = 1.5) +
ggplot2::geom_line(ggplot2::aes(x = publication_year, y = p_cum_technology_vision / sec_axi_fact, color = "Cumulative proportion Technology Corpus"), size = 1.5) +
#
ggplot2::scale_color_manual(
values = c(
"Cumulative proportion OA Corpus" = "#1f77b4",
"Cumulative proportion TCA Corpus" = "black",
"Cumulative proportion vision only corpus" = "#2ca02c",
"Cumulative proportion technology only corpus" = "#d62728",
"Cumulative proportion Technology Corpus" = "#9467bd"
)
) +
ggplot2::scale_fill_manual(
values = c("Nmber of publications per year in TCA Corpus" = "lightgrey")
) +
#
ggplot2::scale_x_continuous(breaks = seq(params$temporal_from, 2030, 10)) +
ggplot2::scale_y_continuous(
"Proportion of publications",
sec.axis = ggplot2::sec_axis(~ . * sec_axi_fact, name = "Cumulative proportion") # divide by 100 to scale back the secondary axis
) +
ggplot2::labs(
title = "Publications over time",
x = "Year",
y = "Number of publications"
) +
ggplot2::theme_minimal() +
ggplot2::theme(
axis.text.y.right = ggplot2::element_text(color = "red"),
legend.position = "inside", # Move the legend to the top left position
legend.justification = c(0.1, 0.9), # Justify the legend to the top left position
legend.background = ggplot2::element_rect(fill = "white", color = "black") # Add a white background to the legend
)
ggplot2::ggsave(
paste0(params$fig_publications_over_time, ".pdf"),
width = 12,
height = 12,
figure
)
ggplot2::ggsave(
paste0(params$fig_publications_over_time, ".png"),
width = 12,
height = 12,
figure
)
rm(figure, sec_axi_fact)
}Sentiment Analysis
Show the code
read.csv("input/ch2_technology/sent_analysis_technology_results_MD.csv") |>
dplyr::group_by(id) |>
dplyr::slice_min(
order_by = row_number(),
n = 1
) |>
saveRDS(file = params$fn_sentiment_results)Spatial
Show the code
#|
if (!file.exists(params$fn_sentiment_spatial_data)) {
data <- corpus_read(params$corpus_authors_dir) |>
dplyr::select(
work_id,
institution_country_code
) |>
dplyr::filter(
!is.na(institution_country_code)
) |>
collect() |>
mutate(
iso3c = countrycode::countrycode(
institution_country_code,
origin = "iso2c",
destination = "iso3c"
),
institution_country_code = NULL
) |>
dplyr::left_join(
readRDS(params$fn_sentiment_results) |>
dplyr::select(
work_id = id,
neg,
neu,
pos,
compound
) ,
by = "work_id"
) |>
dplyr::group_by(iso3c) |>
dplyr::summarize(
mean_neg = mean(neg, na.rm = TRUE),
mean_neu = mean(neu, na.rm = TRUE),
mean_pos = mean(pos, na.rm = TRUE),
mean_compound = mean(compound, na.rm = TRUE),
n = n()
) |>
dplyr::arrange(
dplyr::desc(mean_neg)
) |>
# Filter out missing countries - only necessary as analysis not completed yet
dplyr::filter(
!is.nan(mean_neg)
) |>
saveRDS(file = params$fn_sentiment_spatial_data)
}Show the code
#|
if (length(list.files(path = file.path("maps", "ch2_technology"), pattern = "sentiment_neu_per_countries")) < 4) {
data <- readRDS(params$fn_sentiment_spatial_data)
map <- data |>
map_country_codes(
map_type = "countries",
values = "mean_neu",
geodata_path = params$gdm_dir
) +
ggplot2::scale_fill_gradient2(low = "#E69F00", mid = "white", high = "#56B4E9", midpoint = 0) +
ggplot2::ggtitle("Mean neutral sentiment (0 - 1) - all countries")
map_sel <- data |>
dplyr::filter(n > params$min_count_sentiment_timeseries) |>
map_country_codes(
map_type = "countries",
values = "mean_neu",
geodata_path = params$gdm_dir
) +
ggplot2::scale_fill_gradient2(low = "#E69F00", mid = "white", high = "#56B4E9", midpoint = 0) +
ggplot2::ggtitle("Mean neutral sentiment (0 - 1) - more than 10 works")
ggplot2::ggsave(
file.path("maps", "ch2_technology", "sentiment_neu_per_countries_all.pdf"),
width = 12,
height = 8,
map
)
ggplot2::ggsave(
file.path("maps", "ch2_technology", "sentiment_neu_per_countries_all.png"),
width = 12,
height = 8,
map
)
ggplot2::ggsave(
file.path("maps", "ch2_technology", "sentiment_neu_per_countries_10.pdf"),
width = 12,
height = 8,
map_sel
)
ggplot2::ggsave(
file.path("maps", "ch2_technology", "sentiment_neu_per_countries_10.png"),
width = 12,
height = 8,
map_sel
)
rm(map, data)
}Show the code
#|
if (length(list.files(path = file.path("maps", "ch2_technology"), pattern = "sentiment_pos_per_countries")) < 4) {
data <- readRDS(params$fn_sentiment_spatial_data)
map <- data |>
map_country_codes(
map_type = "countries",
values = "mean_pos",
geodata_path = params$gdm_dir
) +
ggplot2::scale_fill_gradient2(low = "#E69F00", mid = "white", high = "#56B4E9", midpoint = 0) +
ggplot2::ggtitle("Mean positive sentiment (0 - 1) - all countries")
map_sel <- data |>
dplyr::filter(n > params$min_count_sentiment_timeseries) |>
map_country_codes(
map_type = "countries",
values = "mean_pos",
geodata_path = params$gdm_dir
) +
ggplot2::scale_fill_gradient2(low = "#E69F00", mid = "white", high = "#56B4E9", midpoint = 0) +
ggplot2::ggtitle("Mean positive sentiment (0 - 1) - more than 10 works")
ggplot2::ggsave(
file.path("maps", "ch2_technology", "sentiment_pos_per_countries_all.pdf"),
width = 12,
height = 8,
map
)
ggplot2::ggsave(
file.path("maps", "ch2_technology", "sentiment_pos_per_countries_all.png"),
width = 12,
height = 8,
map
)
ggplot2::ggsave(
file.path("maps", "ch2_technology", "sentiment_pos_per_countries_10.pdf"),
width = 12,
height = 8,
map_sel
)
ggplot2::ggsave(
file.path("maps", "ch2_technology", "sentiment_pos_per_countries_10.png"),
width = 12,
height = 8,
map_sel
)
rm(map, data)
}Show the code
#|
if (length(list.files(path = file.path("maps", "ch2_technology"), pattern = "sentiment_neg_per_countries")) < 4) {
data <- readRDS(params$fn_sentiment_spatial_data)
map <- data |>
map_country_codes(
map_type = "countries",
values = "mean_neg",
geodata_path = params$gdm_dir
) +
ggplot2::scale_fill_gradient2(low = "#E69F00", mid = "white", high = "#56B4E9", midpoint = 0) +
ggplot2::ggtitle("Mean negative sentiment (0 - 1) - all countries")
map_sel <- data |>
dplyr::filter(n > params$min_count_sentiment_timeseries) |>
map_country_codes(
map_type = "countries",
values = "mean_pos",
geodata_path = params$gdm_dir
) +
ggplot2::scale_fill_gradient2(low = "#E69F00", mid = "white", high = "#56B4E9", midpoint = 0) +
ggplot2::ggtitle("Mean negative sentiment (0 - 1) - more than 10 works")
ggplot2::ggsave(
file.path("maps", "ch2_technology", "sentiment_neg_per_countries_all.pdf"),
width = 12,
height = 8,
map
)
ggplot2::ggsave(
file.path("maps", "ch2_technology", "sentiment_neg_per_countries_all.png"),
width = 12,
height = 8,
map
)
ggplot2::ggsave(
file.path("maps", "ch2_technology", "sentiment_neg_per_countries_10.pdf"),
width = 12,
height = 8,
map_sel
)
ggplot2::ggsave(
file.path("maps", "ch2_technology", "sentiment_neg_per_countries_10.png"),
width = 12,
height = 8,
map_sel
)
rm(map, data)
}Show the code
#|
if (length(list.files(path = file.path("maps", "ch2_technology"), pattern = "sentiment_comp_per_countries")) < 4) {
data <- readRDS(params$fn_sentiment_spatial_data)
map <- data |>
map_country_codes(
map_type = "countries",
values = "mean_compound",
geodata_path = params$gdm_dir
) +
ggplot2::scale_fill_gradient2(low = "#E69F00", mid = "white", high = "#56B4E9", midpoint = 0) +
ggplot2::ggtitle("Mean compound sentiment (-1: negative; 1: positive) - all countries")
map_sel <- data |>
dplyr::filter(n > params$min_count_sentiment_timeseries) |>
map_country_codes(
map_type = "countries",
values = "mean_compound",
geodata_path = params$gdm_dir
) +
ggplot2::scale_fill_gradient2(low = "#E69F00", mid = "white", high = "#56B4E9", midpoint = 0) +
ggplot2::ggtitle("Mean compound sentiment (-1: negative; 1: positive) - more than 10 works")
ggplot2::ggsave(
file.path("maps", "ch2_technology", "sentiment_comp_per_countries_all.pdf"),
width = 12,
height = 8,
map
)
ggplot2::ggsave(
file.path("maps", "ch2_technology", "sentiment_comp_per_countries_all.png"),
width = 12,
height = 8,
map
)
ggplot2::ggsave(
file.path("maps", "ch2_technology", "sentiment_comp_per_countries_10.pdf"),
width = 12,
height = 8,
map_sel
)
ggplot2::ggsave(
file.path("maps", "ch2_technology", "sentiment_comp_per_countries_10.png"),
width = 12,
height = 8,
map_sel
)
rm(map, data)
}Timeseries
Show the code
#|
if (!file.exists(params$fn_sentiment_temporal_data)) {
data <- readRDS(params$fn_sentiment_results) |>
select(
work_id = id,
year = date,
neg,
neu,
pos,
compound
) |>
dplyr::group_by(year) |>
dplyr::summarize(
neg = mean(neg),
neu = mean(neu),
pos = mean(pos),
compound = mean(compound),
n = n()
) |>
saveRDS(file = params$fn_sentiment_temporal_data)
}Show the code
#|
if (!file.exists(params$fn_sentiment_marine_temporal_data)) {
data <- readRDS(params$fn_sentiment_results) |>
select(
work_id = id,
year = date,
neg,
neu,
pos,
compound
) |>
dplyr::filter(
work_id %in% oa_count$vision_technology_marine_ids
) |>
dplyr::group_by(year) |>
dplyr::summarize(
neg = mean(neg),
neu = mean(neu),
pos = mean(pos),
compound = mean(compound),
n = n()
) |>
saveRDS(file = params$fn_sentiment_marine_temporal_data)
}Show the code
#|
if (length(list.files(path = file.path("figures", "ch2_technology"), pattern = "sentiments_over_time")) < 2) {
figure <- readRDS(params$fn_sentiment_temporal_data) |>
dplyr::filter(
n > params$min_count_sentiment_timeseries
) |>
tidyr::pivot_longer(cols = c(neg, neu, pos, compound), names_to = "type", values_to = "value") |>
ggplot2::ggplot() +
ggplot2::geom_line(aes(x = year, y = value, color = type, linetype = type)) +
ggplot2::scale_color_manual(values = c("black", "red", "blue", "green")) +
ggplot2::labs(
title = paste0("Sentiment Analysis Scores (n > ", params$min_count_sentiment_timeseries, ")"),
x = "Year",
y = "Score",
color = "Type",
linetype = "Type"
) +
ggplot2::theme_minimal()
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_over_time.pdf"),
width = 12,
height = 6,
figure
)
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_over_time.png"),
width = 12,
height = 6,
figure
)
rm(figure)
}Show the code
#|
if (length(list.files(path = file.path("figures", "ch2_technology"), pattern = "sentiments_over_time_neg_pos")) < 2) {
figure <- readRDS(params$fn_sentiment_temporal_data) |>
dplyr::filter(
n > params$min_count_sentiment_timeseries
) |>
tidyr::pivot_longer(cols = c(neg, pos), names_to = "type", values_to = "value") |>
ggplot2::ggplot() +
ggplot2::geom_line(aes(x = year, y = value, color = type, linetype = type)) +
ggplot2::scale_color_manual(values = c("black", "red", "blue", "green")) +
ggplot2::labs(
title = paste0("Sentiment Analysis Scores (n > ", params$min_count_sentiment_timeseries, ")"),
x = "Year",
y = "Score",
color = "Type",
linetype = "Type"
) +
ggplot2::theme_minimal()
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_over_time_neg_pos.pdf"),
width = 12,
height = 6,
figure
)
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_over_time_neg_pos.png"),
width = 12,
height = 6,
figure
)
rm(figure)
}Show the code
#|
if (length(list.files(path = file.path("figures", "ch2_technology"), pattern = "sentiments_marine_over_time_neg_pos")) < 2) {
figure <- readRDS(params$fn_sentiment_marine_temporal_data) |>
dplyr::filter(
n > params$min_count_sentiment_timeseries
) |>
tidyr::pivot_longer(cols = c(neg, pos), names_to = "type", values_to = "value") |>
ggplot2::ggplot() +
ggplot2::geom_line(aes(x = year, y = value, color = type, linetype = type)) +
ggplot2::scale_color_manual(values = c("black", "red", "blue", "green")) +
ggplot2::labs(
title = paste0("Sentiment Analysis Scores (n > ", params$min_count_sentiment_timeseries, ")"),
x = "Year",
y = "Score",
color = "Type",
linetype = "Type"
) +
ggplot2::theme_minimal()
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_marine_over_time_neg_pos.pdf"),
width = 12,
height = 6,
figure
)
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_marine_over_time_neg_pos.png"),
width = 12,
height = 6,
figure
)
rm(figure)
}Show the code
#|
if (length(list.files(path = file.path("figures", "ch2_technology"), pattern = "sentiments_neg_over_time")) < 2) {
figure <- readRDS(params$fn_sentiment_temporal_data) |>
dplyr::filter(
n > params$min_count_sentiment_timeseries
) |>
ggplot2::ggplot() +
ggplot2::geom_line(ggplot2::aes(x = year, y = neg)) +
ggplot2::labs(
title = paste0("Sentiment Analysis negative Score (n > ", params$min_count_sentiment_timeseries, ")"),
x = "Year",
y = "Negative score"
) +
ggplot2::theme_minimal()
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_neg_over_time.pdf"),
width = 12,
height = 6,
figure
)
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_neg_over_time.png"),
width = 12,
height = 6,
figure
)
rm(figure)
}Show the code
#|
if (length(list.files(path = file.path("figures", "ch2_technology"), pattern = "sentiments_neu_over_time")) < 2) {
figure <- readRDS(params$fn_sentiment_temporal_data) |>
dplyr::filter(
n > params$min_count_sentiment_timeseries
) |>
ggplot2::ggplot() +
ggplot2::geom_line(ggplot2::aes(x = year, y = neu)) +
ggplot2::labs(
title = paste0("Sentiment Analysis neutral Score (n > ", params$min_count_sentiment_timeseries, ")"),
x = "Year",
y = "Neutral score"
) +
ggplot2::theme_minimal()
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_neu_over_time.pdf"),
width = 12,
height = 6,
figure
)
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_neu_over_time.png"),
width = 12,
height = 6,
figure
)
rm(figure)
}Show the code
#|
if (length(list.files(path = file.path("figures", "ch2_technology"), pattern = "sentiments_pos_over_time")) < 2) {
figure <- readRDS(params$fn_sentiment_temporal_data) |>
dplyr::filter(
n > params$min_count_sentiment_timeseries
) |>
ggplot2::ggplot() +
ggplot2::geom_line(ggplot2::aes(x = year, y = pos)) +
ggplot2::labs(
title = paste0("Sentiment Analysis positive Score (n > ", params$min_count_sentiment_timeseries, ")"),
x = "Year",
y = "Positive score"
) +
ggplot2::theme_minimal()
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_pos_over_time.pdf"),
width = 12,
height = 6,
figure
)
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_pos_over_time.png"),
width = 12,
height = 6,
figure
)
rm(figure)
}Show the code
#|
if (length(list.files(path = file.path("figures", "ch2_technology"), pattern = "sentiments_comp_over_time")) < 2) {
figure <- readRDS(params$fn_sentiment_temporal_data) |>
dplyr::filter(
n > params$min_count_sentiment_timeseries
) |>
ggplot2::ggplot() +
ggplot2::geom_line(ggplot2::aes(x = year, y = compound)) +
ggplot2::labs(
title = paste0("Sentiment Analysis Compound Score (n > ", params$min_count_sentiment_timeseries, ")"),
x = "Year",
y = "Compound score"
) +
ggplot2::theme_minimal()
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_comp_over_time.pdf"),
width = 12,
height = 6,
figure
)
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_comp_over_time.png"),
width = 12,
height = 6,
figure
)
rm(figure)
}Show the code
#|
if (length(list.files(path = file.path("figures", "ch2_technology"), pattern = "sentiments_marine_over_time_neg_pos_ridge")) < 2) {
figure <- readRDS(params$fn_sentiment_results) |>
select(
work_id = id,
year = date,
neg,
neu,
pos,
compound
) |>
dplyr::ungroup() |>
dplyr::filter(
year >= 1980
) |>
mutate(
year = as.factor(year)
) |>
ggplot() +
geom_density_ridges(aes(x = pos, y = year, fill = "positive"), rel_min_height = 0.005) +
geom_density_ridges(aes(x = -neg, y = year, fill = "negative"), rel_min_height = 0.005)
# Calculate the number of points per year
counts <- figure$data %>%
group_by(year) %>%
summarise(
n = n()
)
# Add the number of points per year as text
figure <- figure +
geom_text(data = counts, aes(x = Inf, y = year, label = n), hjust = "inward", vjust = 0.5)
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_marine_over_time_neg_pos_ridge.pdf"),
width = 12,
height = 6,
figure
)
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_marine_over_time_neg_pos_ridge.png"),
width = 12,
height = 6,
figure
)
rm(figure)
}Show the code
#|
if (length(list.files(path = file.path("figures", "ch2_technology"), pattern = "sentiments_over_time_neg_pos")) < 2) {
figure <- readRDS(params$fn_sentiment_temporal_data) |>
dplyr::filter(
n > params$min_count_sentiment_timeseries
) |>
tidyr::pivot_longer(cols = c(neg, pos), names_to = "type", values_to = "value") |>
ggplot2::ggplot() +
ggplot2::geom_line(aes(x = year, y = value, color = type, linetype = type)) +
ggplot2::scale_color_manual(values = c("black", "red", "blue", "green")) +
ggplot2::labs(
title = paste0("Sentiment Analysis Scores (n > ", params$min_count_sentiment_timeseries, ")"),
x = "Year",
y = "Score",
color = "Type",
linetype = "Type"
) +
ggplot2::theme_minimal()
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_over_time_neg_pos.pdf"),
width = 12,
height = 6,
figure
)
ggplot2::ggsave(
file.path("figures", "ch2_technology", "sentiments_over_time_neg_pos.png"),
width = 12,
height = 6,
figure
)
rm(figure)
}Extract `marine papers with sentiment scores
Show the code
#|
if (!file.exists(params$fn_marine_sentiment_all)) {
corpus_read(params$corpus_dir) |>
dplyr::filter(
id %in% oa_count$vision_technology_marine_ids
) |>
dplyr::select(id, doi, author_abbr, display_name, ab) |>
dplyr::rename(abstract = ab, title = display_name) |>
dplyr::collect() |>
dplyr::slice_sample(n = 250) |>
dplyr::mutate(
abstract = substr(abstract, start = 1, stop = 5000)
) |>
dplyr::left_join(
y = readRDS(params$fn_sentiment_results) |>
dplyr::select(
id,
neg,
neu,
pos,
compound
),
by = "id"
) |>
writexl::write_xlsx(path = params$fn_marine_sentiment_all)
}
if (!file.exists(params$fn_marine_sentiment)) {
corpus_read(params$corpus_dir) |>
dplyr::filter(
id %in% oa_count$vision_technology_marine_ids
) |>
dplyr::select(id, doi, author_abbr, display_name, ab) |>
dplyr::rename(abstract = ab, title = display_name) |>
dplyr::collect() |>
dplyr::slice_sample(n = 250) |>
dplyr::mutate(
abstract = substr(abstract, start = 1, stop = 5000)
) |>
dplyr::left_join(
y = readRDS(params$fn_sentiment_results) |>
dplyr::select(
id,
neg,
neu,
pos,
compound
),
by = "id"
) |>
writexl::write_xlsx(path = params$fn_marine_sentiment)
}Results
The results are based on data downloaded or accessed at:
Show the code
c_time <- list.files(
path = params$pages_dir,
recursive = TRUE,
pattern = ".rds$",
full.names = TRUE
) |>
file.mtime() |>
as.Date() |>
unique()- Technology Corpus: located at
./data/ch2_technology/corpusdownloaded at 2024-03-27 from OpenAlex - Counts: obtained from at 2024-04-25 from OpenAlex
vision AND technology in TCA Corpus
For the TCA Corpus, we do have 490,201 number of works.
Random Sample of 250 works
An Excel file conataining a random sample of 250 works from the Technology Corpus (technology AND vision AND nature AND transformativechange) with the fields id, doi, author_abbr and abstract of the papers. The Excel file can be downloaded from here.
Subfields
The subfields are based on the main topic assigned to each work. There are other topics also assigned, but this one has been identified as the main topic by an algorythm. count is the number of works in the vision AND technology corpus which have been assigned to the subfield.
Please take a look at these subfields of the topics to identify the ones to be filtered out.
The easies would be to download the Excel file through the button and to mark the subfields to be filtered out.
Show the code
IPBES.R::table_dt((oa_count$vision_technology_subfields |> dplyr::arrange(desc(count))), fixedColumns = NULL, fn = "Vision Technology Subfields")Publications over time
A pdf of the graph can be downloaded here.
Sentiment Analysis
Two .parquet files containing the id, publication_year and ab (abstract) were extracted and are available upon request due to their size.
For analyzing the sentiments of the provided abstracts, we have used the Python NLTK package, and VADER (Valence Aware Dictionary for Sentiment Reasoning) which is an NLTK module that provides sentiment scores based on the words used. VADER is a pre-trained, rule-based sentiment analysis model in which the terms are generally labeled as per their semantic orientation as either positive or negative.
The main advantage/reason for using this model was that it doesn’t require a labbed training dataset. The output of the model is 4 statistical scores:
- compound: composite score that summarizes the overall sentiment of the text, where scores close to 1 indicate a positive sentiment, scores close to -1 indicate a negative sentiment, and scores close to 0 indicate a neutral sentiment
- negative: percentage of negative sentiments in the text
- neutral: percentage of neutral sentiments in the text
- positive: percentage of positive sentiments in the text
Show the code
#|
readRDS(params$fn_sentiment_results) |>
IPBES.R::table_dt(fn = "sentiment_scores", fixedColumns = list(leftColumns = 2))Warning in instance$preRenderHook(instance): It seems your data is too big for
client-side DataTables. You may consider server-side processing:
https://rstudio.github.io/DT/server.html